by M. Ali Yikilmaz
3 April 2026
PhD research in mechanical engineering increasingly benefits from the integration of artificial intelligence, not as a replacement for the research process, but as a means of enhancing it. In the context of advanced manufacturing, AI is being utilized to support sensor-based monitoring approaches that aim to detect defects during metal 3D printing. These manufacturing processes generate large volumes of complex data from thermal and optical imaging systems, and making sense of this information in real time is essential for ensuring process stability and product quality.
Rather than serving as a substitute for critical scientific thinking, AI tools are used to refine, validate, and accelerate researcher-driven ideas. Literature review, experimental design, and conceptual development remain foundational components of the research process. Interactive AI platforms can assist in tasks such as exploring coding strategies or clarifying technical topics, but the scientific direction remains firmly grounded in human expertise and judgment.
More broadly, machine learning and deep learning are opening new possibilities in engineering by enabling data-driven insights from complex multi-sensor datasets. In additive manufacturing, these methods are increasingly used to analyze melt pool behavior, identify spatter events, and detect subtle anomalies that might otherwise be overlooked. Techniques such as convolutional neural networks (CNNs) and vision transformers (ViT) enable image-based defect detection, while autoencoders facilitate unsupervised anomaly identification, and clustering algorithms help reveal relationships between process parameters and part quality. These and many other deep learning algorithms are currently being employed to ensure part quality, enhance process understanding, and support the development of more reliable and adaptive manufacturing systems.
These tools serve as amplifiers of scientific capability. They enable the transformation of raw, high-dimensional data into actionable knowledge and support the investigation of patterns too complex for traditional methods alone. Their thoughtful integration into research workflows has made the engineering research process more dynamic, data-informed, and capable of addressing emerging challenges in advanced manufacturing.